关键词: 火焰识别, 多特征量, 对数回归, 嵌入式视频, 实时火灾预警

Abstract: In order to improve the recognition rate and reduce the false-recognition rate in real-time detection of flame in video surveillance, a fast flame detection algorithm based on multi features logarithm regression model is proposed in this paper. The image is segmented according to the chromaticity of the flame, and the candidate fire region (CFR) is obtained by subtraction the moving target image with reference image. Then the features of the CRF such as area change rate, circularity, number of sharp corners and centroid displacement are extracted to establish the logarithmic regression model. A total of 300 images including flame and non-flame images, which are got from NIST, VisiFire, ICV and experimental libraries are used to train parametric learning. Then 8 video clips including 11071 images are used to validate the proposed algorithm. The experimental results show that the true positive rate (TPR) and true negative rate (TNR) are 93% and 98% compared with the other methods, respectively., Therefore, it provides the precondition of the next embedded system for fast fire recognition.

Key words: Flame reorganization, Multi features, Longitude regression, Embedded video, Real-time fire alarm